Surg Clin N Am 85 (2005) 11911213

The Management of Postoperative Bleeding

T. Forcht Dagi, MD, MPH
Division of Health Sciences and Technology, The Harvard-MIT Program in Health Sciences and Technology, 423 Commonwealth Avenue, Newton Center, MA 02459, USA

This article addresses the management of postoperative bleeding. The problem is called postoperative bleeding rather than postoperative hemorrhage to emphasize the fact that perfect postoperative hemostasis rather than acceptable postoperative blood loss is the ideal. Postoperative bleeding is a risk of all surgical procedures. The best way to reduce the risk of hemorrhage is to identify and correct potential causes of coagulopathy preoperatively as well as postoperatively. In the presence of bright red bleeding from any site, if the prothrombin time (PT), activated partial thromboplastin time (aPTT), platelet count, and temperature are normal, urgent re-exploration is indicated unless other factors dictate a more thorough diagnostic workup. In the presence of lifethreatening hemorrhage, control of bleeding takes priority. The hematologic workup is pursued in parallel. Throughout this article, a distinction is made between technical causes of bleeding and coagulopathy, or disorders of hemostasis [1]. The term technical causes of bleeding refers to four broad categories of postoperative blood loss: 1. Inadequate repair of vessels or vascular structures that are knowingly opened or divided, whether purposefully or accidentally 2. Occult or undiagnosed and therefore unrepaired injury to the vascular system

In writing this article, the author has drawn liberally from the several authors cited, particularly references 1 and 14. In several instances the author has paraphrased and combined the authors and followed their outline closely, with only small changes in language and order of presentation for purposes of flow and to make the material more pertinent to the perioperative setting and to this publication. E-mail address: tdagi@post.harvard.edu 0039-6109/05/$ - see front matter 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.suc.2005.10.013 surgical.theclinics.com

1192

DAGI

3. Injury or damage during the course of surgery to organs or structures within the operative eld, whether recognized at the time or not 4. Injury or damage during the course of surgery or in the immediate postoperative period to organs or structures remote from the surgical site Except in the most general sense, this article does not cover the management of intraoperative or postoperative hemorrhage arising from technical causes, the management of anticoagulation in the perioperative period, the management of trauma, the management of transfusion, or the treatment of shock. Few surgeons have the expertise in hematology or the time to manage the medical aspects of coagulopathy without consultation. This article provides the surgeon with a basis on which to engage, rather than to replace, the medical specialist. Early consultation may be prudent and useful. Preoperative screening strategies Routine preoperative screening In terms of surgical complications, an ounce of prevention is worth a pound of cure, and preoperative screening has evolved with the thought of preventing intra- and postoperative problems. What constitutes a reasonable stratagem for preoperative screening for hemostasis? The guidelines of the Joint Commission on Accreditation of Health care Organizations mandate a preoperative evaluation to assess a patients readiness and risk for surgical intervention but leave the details up to each individual organization [2]. As increasingly sophisticated screening tests have entered the market, their cost and their aggregate eectiveness and utility for routine use have come into question. Unlike their quantiable value in elucidating the cause of demonstrated coagulopathy, their screening value, as measured in terms of (1) preventing postoperative bleeding and (2) the costs of sorting out abnormalities of questionable signicance, remains in question [3,4]. The signicance of abnormal test results Test results may be reported on a continuum (ie, a set of values between 0 and innity, as in a patients weight), on an ordinal scale with a small number of discrete and discontinuous values (ie, 1 , 2 , and so forth, as with dipstick examination of proteinuria), or categorically (eg. as normal or abnormal). Tests that report results on a continuum must provide cuto points for abnormally high or low values. The intervening values, classically obtained by calculating the mean and allowing 2 standard deviation on either side, typically represent the range of results in 95% of a reference population. As a result, 5% of any population comparable to the reference population would be expected to have an abnormal result on any given test, and this probability increases when multiple tests are ordered. In

MANAGEMENT OF POSTOPERATIVE BLEEDING

1193

a patient who has no disease, there is a 64% likelihood of an abnormality being found on a chemistry panel of 20 tests [4]. Nevertheless, for medico-legal reasons, to detect unsuspected and potentially correctable abnormalities that increase surgical risk, to establish a baseline value for laboratory values that might change postoperatively or need to be monitored because of perioperative medications (eg, anticoagulants), or because of the importance of diagnosing related conditions (such as carotid stenosis in the face of symptomatic coronary artery disease), routine preoperative screening is standard of care. It has been noted that clinicians ignore between 30% and 60% of abnormalities elicited on routine screening [46]. Ignoring abnormalities without comment increases the risk of suit. Unless the chart contains good documentation reecting the reasons for setting aside an abnormal screening result, the physician is at risk for a plaintis verdict in the event of legal action following a surgical complication or adverse result [4]. The term screening is properly reserved for patients who do not have signs and symptoms of underlying abnormality and who are free of any known conditions that increase the likelihood of abnormal results. Screening tests are useful when they 1. 2. 3. 4. 5. 6. 7. Cost little Are consistent Carry negligible risk Demonstrate high sensitivity and selectivity Oer high positive and negative predictive value Accurately foreshadow surgical morbidity Are appropriate to the population at hand (eg, there is questionable value to skin testing for tuberculosis in a population known to have been vaccinated with Bacille Calmette-Guerin [BCG]) 8. Uncover common conditions contributing to surgical morbidity and for which eective intervention is available [4]

Thus, to be ecient and eective, screening stratagems must be useful and pertinent. The literature cautions that estimates of unsuspected abnormalities in healthy populations are probably exaggerated and that routine preoperative screening is of limited measurable value unless abnormal results are suspected, or abnormal results have been obtained previously [4,7,8]. Nevertheless, the risks accompanying surgical intervention, both to the patient and to the surgeon, justify judicious screening to detect and correct pertinent underlying abnormalities. History and physical examination Patients who do not have a personal or family history of bleeding diculties or abnormal bleeding associated with dental extractions, previous surgery, routine childhood and adolescent trauma, or childbirth are unlikely

1194

DAGI

to suer from familial or congenital coagulopathy. Patients taking no medications and without history of bleeding disorders are also at very low risk. Many patients, however, do not recognize that over-the countermedicationsdnonsteroidal anti-inammatory drugs most notablyddo count as medications and can interfere with normal coagulation. The history must specically address the casual use of over-the-counter drugs as well as the routine and prescribed use of any medications. The absence of bruising or other signs of bleeding on routine physical examination helps conrm that the patient is at low risk for surgical bleeding. This nding should be clearly listed in the physical examination. Abnormalities in routine screening and their signicance The following statistics refer to only the results of tests performed for screening purposes in the context described previously. Complete blood cell count The overall incidence of hemoglobin abnormalities in a combined review of 9363 patients was 1.8%, but anemia occurs in 4% to 9% of patients 70 years of age and older (n 526) and predicts the need for transfusion in patients at risk for blood loss [4,911]. The prevalence of an elevated white blood cell count in a combined review of 5359 patients was less than 1% and was unrelated to perioperative morbidity [4,11,12]. The prevalence of platelet abnormalities in a combined review of 8670 patients was calculated to be 0.9% [4,12]. The abnormality in the preponderance of cases was thrombocytopenia, but management was changed in only 0.02% of cases [4]. Coagulation The yield for abnormal PT in a combined review of 3786 patients was 0.3%, but in no case was management inuenced [4]. Evidence suggests that the PT alone is a poor screening test and neither predicts nor excludes clinically relevant perioperative bleeding abnormalities [10]. The yield for abnormalities in PTT was 6.5% in a combined review of 2955 patients [4]. These gures are summarized in Table 1. Bleeding time The use of bleeding time for screening is controversial [13]. On the one hand, there is good evidence to suggest the actual times are variable and depend on technique. It has been asserted that bleeding time oers no screening advantage in normal reference populations, and therefore it has no role to play [4,13,14]. On the other hand, bleeding time is a good way to detect abnormalities of platelet function. Disruption of platelet function is increasingly common in the aging population because of the regular use of aspirin (ASA) for primary

MANAGEMENT OF POSTOPERATIVE BLEEDING

1195

Table 1 Prevalence of hematological abnormalities in preoperative screening Test Hemoglobin Leukocytes Platelets Prothrombin time Partial thromboplastin time Number of Patients 9363 5359 8670 4786 2955 Prevalence of Abnormality (%) 1.8% 0.7% 0.9% 0.3% 6.5%

and secondary prophylaxis in ischemic heart disease and cerebrovascular occlusive disease, the occasional and often unreported use of ASA for pain and inammation, and the use of ticlopidine hydrochloride or clopidogrel bisulfate for platelet inhibition in cerebrovascular disease (where ASA has failed) or for stenting in coronary artery disease (clopidogrel and ASA in combination) [14]. Bleeding times normalize after these drugs have been discontinued, but the time required for normalization cannot be predicted with certainty [14]. Eptibatide, abciximab, and tiroban hydrochloride block the integrin aIIbb3 receptor on the platelet membrane. Blocking this receptor paralyzes the platelet and prevents activation. Platelets fail to aggregate normally, and their endothelial attachments do not withstand high shear forces. Abciximab has additional eects as well [15]. Bleeding time typically returns to normal 6 hours after the discontinuation of eptibatide but more than 24 hours after the discontinuation of abciximab. Abciximab is also associated with thrombocytopenia, generally within the rst 24 hours of use. Platelet counts drop below 100,000 in 6% of patients and below 50,000 in 1.5% [15]. The prevalence of thrombocytopenia is lower with tiroban and apparently is nonexistent with eptibatide [14]. Thrombocytopenia has also been associated with the use of clopidogrel [16]. Statistics for the prevalence of abnormalities of bleeding time in screening of a reference population are not available, perhaps because bleeding times are not routinely measured. Even so, the measurement of bleeding time is indicated in patients treated with platelet-inhibiting agents, irrespective of symptomatology or ndings, and in patients who are likely to self-medicate with ASA, wittingly or not. (Many patients and some physicians are unaware that ASA may be found in combination with antihistamines and decongestants in cold and headache remedies, and, in some parts of the United States, packaged as a powder or as a branded pill without ASA in its name.) Abnormalities of bleeding time do represent a coagulopathy and must be respected as such. They should not be ignored. In some cases it is easier to wait until platelet function returns to normal. In other cases, platelet transfusions are in order. Plasma exchange may be indicated for drug-induced thrombocytopenia purpura [16].

1196

DAGI

Surgery in higher-risk populations Sometimes it is necessary to proceed with surgery urgently despite the likelihood of defective hemostasis and intra- or postoperative bleeding. In these cases, preoperative screening has three purposes: to identify defects in hemostasis that can be corrected preoperatively, to guide the management of hemostatic defects that cannot be corrected in the time remaining before surgery, and to help manage the bleeding that cannot be prevented. A perspective on risk management and benets of screening Increasing emphasis on cost containment in medicine has resulted in attempts to distinguish essential from nonessential measures in patient care. This trend has been balanced, at least in part, by concerns for patient safety and fears of medical liability actions. Even the terms used create difculty: where does advisable or prudent t in the spectrum of nonessential to essential? The literature on preoperative screening reports a low prevalence of abnormal ndings in study populations and an even lower likelihood that management would be altered by any single discovered abnormality. In populations at dierent and usually greater risk, such as patients treated with low-dose ASA after transient ischemic attack, the prevalence of abnormality will be greater if the appropriate test is performed. Bleeding time may not be indicated as a screening test in the population at large, but bleeding time may be very useful to screen for coagulopathy in patients whose platelet function is inhibited for therapeutic purposes or in patients likely to self-medicate unwittingly with platelet-inhibiting preparations. What is the real signicance of these observations? Given the large number of surgical procedures performed every day, a 0.01 prevalence of abnormality leading to even a 0.00001 likelihood of alteration in management aects a large aggregate number of patients and surgeons in the aggregate. The aggregate risk of surgical bleeding in a particular population is derived by combining the risk of bleeding in the study population with an estimate of the risk of technical causes of bleeding associated with the procedure. It is reasonable to attempt to identify situations in which both the populationbased risk of bleeding and the technical risk are exceptionally low and to raise the threshold for screening accordingly. One thoroughly studied example is cataract surgery in patients who have absolutely no known or discoverable risk factors for bleeding, for whom routine screening really does seem to convey no measurable benet [17]. In the broader perspective of cost-benet analysis, it is prudent to think not only in terms of the technical risk of surgery but also of the consequences of bleeding. Because of the specic problems associated with intracranial pressure and vasospasm following intracranial hemorrhage, for example, the costbenet analysis of screening in intracranial surgery may shift in comparison to other operations with similar technical risk.

MANAGEMENT OF POSTOPERATIVE BLEEDING

1197

A classication of surgical risk that factors anticipated blood loss into the risk of surgery can serve as a useful guide. A good example is the Johns Hopkins Risk Classication System [3]. It can be used either alone or in combination with other classication systems. The Hopkins system associates category I (minimally invasive) with little or no blood loss, category II (minimally to moderately invasive) with blood loss less than 500 mL, category III (moderately to signicantly invasive) with a blood loss potential of 500 to 1500 mL, and categories IV and V (both considered highly invasive) with blood loss in excess of 1500 mL [3]. Every classication system has inherent limits. Risk analysis is useful for predicting the events in a set (a dened population) but not the behavior of a statistic (a single member of the set) in the abstract. Thus, there is no way to know exactly what will transpire with a particular patient because there is no perfect way to combine the pertinent risks into a perfect prediction for an individual case. For this reason alone, it is often appropriate to move beyond routine screening protocols when there is concern about factors that might jeopardize patient safety, even in category I or category II patients. In the face of such concerns, it may be advisable to move from screening protocols to diagnostic protocols such as those reviewed here and used to rule out technical causes of bleeding. Diagnostic protocols Overview The rst step in diagnosis is recognizing that the patient is bleeding. This recognition is neither as obvious nor as easy as it might seem. Vital signs may remain remarkably stable, especially in the young, until shock ensues. Normal blood pressure may be preserved despite the loss of 1.5 to 2 L, or 40% of total blood volume. Falling hematocrits and hemoglobin levels are often dismissed as artifacts of dilution. Drains placed intraoperatively and intended to monitor and drain blood from a closed cavity may become blocked, kinked, or malpositioned, thereby providing a false sense of security. Finally, the bleeding may be remote from the site of surgery or it may accumulate in an undrained compartment. The results of physical examination should be integrated with all other data sources. Tachycardia, diminished cardiac output, dropping central venous pressure, reductions in urine output, and abnormal capillary rell pattern are all suggestive of bleeding, as are ank bruises and swelling of the extremities with discoloration. The search for an occult technical cause of bleeding should continue, regularly and repeatedly, until the patient is stable and bleeding has stopped. Technical causes of bleeding and coagulopathy may coexist and often do. It may be dicult to address the technical cause before stabilizing the coagulopathy. On the other hand, massive transfusion of blood products may

1198

DAGI

induce or worsen coagulopathy. Both lines of managementdcontrol of technical causes of bleeding and reversal of coagulopathydshould be pursued simultaneously. The precise details of intervention and timing are often a matter of judgment and cannot be reduced eectively to protocol. Intracranial bleeding may lead to neurogenic shock or neurogenic pulmonary edema, but the volume of hemorrhage does not per se reduce circulating blood volume enough to lead to shock except in infants, and even then very rarely. The search for technical sources of hemorrhage includes repeated physical examination and appropriate imaging techniques (CT, MRI, MR angiography, ultrasound). Sometimes angiography is indicated. Angiography should be seen as an opportunity for endovascular repair, if possible and appropriate. Screening protocols versus diagnostic protocols The approach to preoperative screening diers from the approach to the diagnosis of coagulopathy. In the preoperative situation, the patient can be assumed to have normal hemostatic function unless history or ndings dictate otherwise. In the setting of postoperative hemorrhage (and following the exclusion of technical causes of bleeding), the patient is assumed to have a coagulopathy. For preoperative purposes, normal test results up to 4 months old are generally deemed reliable if the patient is clinically unchanged [4,8]. In the postoperative setting, test results can change quickly, and serial studies may be required. The diagnosis of coagulopathy is directed at assessing the function of the factors, cells, and other elements that contribute to normal coagulation. Technical causes of bleeding and coagulopathy may occur simultaneously or sequentially. Until the bleeding has come under control, until it has been shown not to recur, and until the patient is stable, neither the exclusion of one nor the conrmation of the other suces. Assessment of coagulopathy The six initial steps in assessing coagulopathy are to 1. 2. 3. 4. 5. 6. Perform complete blood cell count and coagulation studies Check for and correct hypothermia Review the history Review medications Obtain additional studies if indicated Check serially for new or ongoing sources of bleed loss

One of the most cogent algorithms has been published by Owings and Gosselin in ACS Surgery [1]. The discussion that follows draws liberally on their work. Complete blood cell count and coagulation studies. Blood is drawn and sent in two tubes, one containing ethylenediaminetetraacetic acid [EDTA] for

MANAGEMENT OF POSTOPERATIVE BLEEDING

1199

a complete blood cell count, and the other citrated for coagulation analysis. The tubes must be carefully labeled for identication. It is easy for samples to be confused or mislabeled in the press of resuscitation. The samples should be sent stat. It is the responsibility of the treating physician to follow up on the studies and make sure both the studies and their results are documented in the chart. Hypothermia. A body temperature below 35 C (95 F) may inhibit clotting mechanisms. In pure hypothermic coagulopathy, coagulation mechanisms normalize when normal temperature is restored. Hypothermia may contribute to or be the sole source of coagulopathy. Hypothermic patients are actively rewarmed. Dierential diagnosis of coagulopathy. Most coagulopathies can be diagnosed, or at least separated into categories reecting the most likely mechanism or mechanisms of coagulopathy, on the basis of the international normalized ratio (INR), the aPTT, platelet count, platelet function, and family history. Sometimes, the diagnosis cannot be further rened without assays for specic factors. The number of platelets in circulation and platelet function are independent variables. Bleeding time is the traditional test used to assess platelet function. Its limitations have been noted. Although a number of far more sophisticated tests have been introduced recently, the problem of adequate assessment of platelet dysfunction is beyond the scope of this article. A summary of dierential diagnosis is given in Table 2. A discussion of the INR is given in Appendix 1. The management of postoperative bleeding Overview In the presence of bright red bleeding from any site, if the PT, aPTT, platelet count, and temperature are normal, urgent re-exploration is indicated unless other factors dictate a more thorough diagnostic workup. In the presence of life-threatening arterial hemorrhage, control of bleeding takes priority. The hematologic workup is pursued in parallel. The most basic principles of management are 1. Diagnose and treat shock and any other potentially life-threatening conditions 2. Rule out technical sources of bleeding 3. Restore clotting parameters to normal by means of medications, transfusion of blood products or clotting factors, restoration of normothermia, diagnosis and treatment of sepsis, or control of other precipitating or contributing factors (eg, retained products of conception, as discussed later) 4. Monitor for stability of clotting parameters

Table 2 Dierential diagnosis of postoperative coagulopathies Symptom Persisting ooze, low-volume bleeding INR Normal aPTT Normal Platelet Number Normal Platelet Functiona Impaired Family or Personal History Negative Diagnosis Impaired platelet function Dierential Diagnosis Comments Spontaneous bleeding with thrombocytopenia !50,000 is rare (some authorities cite 20,000 [1], but postoperative needs may be higher

Persisting ooze, low-volume bleeding Persisting ooze, low-volume bleeding Minor or major bleeding Minor or major bleeding Major or minor bleeding, oozing

Normal

Normal

Low

Normal

Negative

Thrombocytopenia

Normal

Normal

Normal

Normal

Negative

Normal

Prolonged

Normal

Normal

Negative

Drug-induced coagulopathy, most likely unfractionated heparin in the US

Consider factor XIII deciency, hypobrinogenemia, dybrinogenemia, altered brinolysis Eects of direct thrombin inhibitors such as hirulog and lepirudin von Willebrands disease Conrm by testing for circulating von Willebrands factor levels Rarely encountered without prior family or personal history. Specic factor analysis is indicated.

Normal

Prolonged slightly Normal

Normal

Abnormal

Negative

Normal

Normal

Normal

Positive for hemophilia

Hemophilia

Hemophilia A: factor VIII deciency. Hemophilia B: factor IX deciency. Hemophilia C: factor XI deciency.

Major or minor bleeding

Increased

Normal

Normal

n/a

Hepatitis, liver disease, ETOH, oral anticoagulation

Factor deciency, vitamin K deciency, warfarin eect

Major bleeding

Prolonged

Prolonged

Normal

Normal

Multiple factor deciencies

Undiagnosed liver disease, malabsorption, malnutrition, warfarin eect or superwarfarin (rat poison) toxicity, antibiotic eect DIC, hemodilution, uremia and nephrotic syndrome Isolated, rare factor deciencies of the common pathway include factors X, V, and prothrombin Autoimmune conditions resulting in acquired factor V deciencies Factor X deciency associated with amyloidosis Acquired hypoprothrombinemia associated in lupus Warfarin overdose Rodenticide toxicity Animal venoms

Reversal of anticoagulation with vitamin K accurately reected in INR

D-dimer level assay helps the diagnosis of DIC. If level ! 1000 ng/mL, DIC unlikely; level O 2000 ng/mL without confounding explanation, DIC highly likely

Abbreviations: aPTT, activated partial thromboplastin time; DIC, diuse intravascular coagulation; INR, international normalized ratio; ETOH, alcohol ingestion. Note: platelet count and platelet function are independent variables. The bleeding time has been used traditionally to assess platelet function. Its drawbacks have been noted. Bleeding time may yield normal results in up to 50% of patients with congenital thrombocytopathies. See standard texts for description of newer tests for platelet function [1].
a

1202

DAGI

5. Correct as needed 6. Monitor for new or ongoing sources of bleeding or blood loss 7. Monitor for treatable complications associated with new or ongoing sources of bleeding or blood loss (eg, intracranial hemorrhage) Generally accepted criteria for transfusion are published widely and are not reviewed here. Table 3 provides a summary of current recommendations for replacement transfusion in acute blood loss [18]. Technical causes The following discussion draws liberally from McKenna [14]. Postoperative bleeding can result from one cause or many, and the causes may be related or linked or not. By default, the surgeon must rst consider a surgically remediable technical cause of bleeding. Technical causes are least likely to respond to nonsurgical intervention. Principles of surgical hemostasis are reviewed in all standard general surgical and specialty textbooks and are not revisited here, except to note that the use of certain materials and techniques in surgery may cause coagulopathy in unexpected ways. For example, high levels of suction have been associated with diuse intravascular coagulation (DIC) after surgery for scoliosis when a cell-saver has been deployed. The development of antibovine antibodies after the use of topical bovine thrombin in neurosurgery has been associated with coagulopathy [14]. Nontechnical causes Hypothermia The eects of hypothermia have already been noted. Hypothermia may be the sole cause of coagulopathy or may contribute to coagulopathy
Table 3 Transfusion Guidelines in Acute Blood Loss Criterion Ischemia Degree of blood loss Hemoglobin concentration Vital signs Signicance High risk of occurrence, impending and/or increased risk to patient because of underlying medical issues: transfusion usually indicated O 30% rapid blood loss: transfusion may be indicated (! 30% rapid blood loss in previously healthy patient: usually well tolerated) ! 6 g/dL: transfusion generally required. 610 g/dL: dictated by clinical circumstances. O 6 g/dL: transfusion rarely required Tachycardia and hypotension refractory to volume expansion with hemoglobin concentration in 610 g/dL range and extent of blood loss unknown: transfusion usually required O2 extraction ratio O 50% with VO2 decrease: transfusion usually required

O2 extraction ratio

From Simon TL, Alverson DC, AuBuchon J, et al. Practice parameters for the use of red blood cell transfusions: developed by the Red Blood Cell Administration Practice Guideline Development Task Force of the College of American Pathologists. Arch Pathol Lab Med 1998;122:1308; with permission.

MANAGEMENT OF POSTOPERATIVE BLEEDING

1203

from other root causes. Laboratories warm samples to 37 C. In pure hypothermic coagulopathy, normal coagulation parameters will be documented. If the patient exhibits normal coagulation parameters and is normothermic, attention should revert to the diagnosis of occult technical causes of bleeding [1]. In instances of induced hypothermia for therapeutic purposes, a thorough riskbenet analysis comparing the risks of coagulopathy with the risks of normothermia should be conducted.

Drugs Aspirin. ASA is the most common platelet-inhibiting drug. The eect is dose related and aects bleeding times. ASA eects may last up to 10 days and cannot be completely reversed directly, unlike those of other nonsteroidal anti-inammatory drugs such as ibuprofen [1]. The eects are potentiated when ASA is given in conjunction with other platelet-function inhibitors or to patients who have other causes of coagulopathy, such as type I von Willebrands disease or hepatic or renal dysfunction. The bleeding time may become very prolonged. The rst line of treatment is platelet transfusion. Desmopressin acetate may be helpful, particularly in patients who have renal failure [19]. Desmopressin acetate may cause water retention, hyponatremia, and, more rarely, arterial thrombotic complications. Platelet inhibitors. The eects of ticlopidine, clopidogrel, eptibatide, abciximab and tiroban have been noted. Prolonged bleeding time may be reversed with platelet transfusions, which are indicated specically when an accompanying thrombocytopenia ensues after the use of abciximab or tiroban. Anticoagulants. Unfractionated heparin (UFH) is commonly administered in large doses to provide systemic anticoagulation in the management of venous thrombosis, hypercoagulability, arterial dissection, and arterial embolic states such as transient ischemic attack or stroke-in-evolution. UFH prolongs the aPTT. In low, prophylactic doses, heparin is used to prevent deep venous thrombosis or for catheter ushing. Even low doses of UFH may cause clinically signicant anticoagulation in the setting of hepatic or renal dysfunction. When the consequences of bleeding or hematoma are not critical, it may be possible to manage UFH-mediated anticoagulation by modifying the dose of UFH. When the consequences are critical (eg, intracranial surgery), the eects must be reversed. The eects of UFH last 24 to 72 hours after administration. If more urgent reversal is required, prolonged aPTT may be reversed by protamine sulfate infusion. Rapid reversal, however, may precipitate unwanted thrombosis. Protamine administration may also lead to allergic reactions and anaphylaxis, and in excess, protamine may actually prolong the aPTT. Sensitivity to protamine may be related to sensitization to insulin [1].

1204

DAGI

The aPTT may be artifactually prolonged when blood is drawn from a heparin-ushed catheter. Conrmation may be obtained by comparison with an aPTT drawn by venipuncture. UFH administration may also induce the formation of heparin platelet antibodies and result in heparin-induced thrombocytopenia (HIT), or, even more rarely and paradoxically, in a life-threatening heparin-induced thrombocytopenia thrombosis syndrome (HITTS). These syndromes develop in approximately 1% to 5% of patients within 3 to 5 days of starting treatment with heparin, and approximately 30% to 40% of patients who have HIT progress to HITTS. The mortality and morbidity associated with HITTS reaches 40% to 55% [20]. The heparinplatelet antibody interactions may explain only one part of HIT. It is a complex phenomenon that remains to be elucidated fully [14]. The importance of HIT and HITTS derives in part from the prevalence of heparin administration. These conditions also exemplify a complex coagulopathy with both hemorrhagic and thrombotic features. The diagnosis may elude the surgeon unless specically pursued. The diagnosis of HIT is made on the basis of a 50% drop in baseline platelet count or thrombocytopenia below 100,000. A 30% drop serves as an indication to assay heparin-platelet antibodies using a combination of the three tests currently available: 14C-serotonin release assay which measures the presence of functionally active heparin-platelet antibody; the platelet factor 4 ELISA, a more sensitive but less specic for identifying heparin-platelet antibodies; the platelet aggregation assay, which helps conrm the diagnosis when abnormal. Other methods to identify heparin-platelet antibodies are in development. Low molecular weight heparins (LMWH) such as enoxaparin sodium, tinzaparin sodium, and dalteparin sodium have greater than a 90% crossreactivity with heparinplatelet antibodies and must not be used as substitutes, even when the immediate threat of postoperative bleeding has past. A full discussion of the management of HIT and HITTS goes beyond the purview of this article, but, very broadly, the recommendation is to treat with warfarin supplemented by recombinant hirudin or argatroban until the dose is stabilized and satisfactory anticoagulation achieved. The anticoagulant eects of LMWHs are best assayed by an anti-factor Xa assay. Levels exceeding 1.3 to 2.0 IU/mL may be deemed abnormally high. The aPTT does not eectively measure the anticoagulant eect of LMWHs. A normal aPTT may still be associated with a bleeding diathesis [14]. In contrast to the eectiveness of protamine administration in reversing anticoagulation with UFH, protamine does not reliably reverse the anticoagulant eect after LMWH administration. Fresh-frozen plasma (FFP) contains antithrombin. Antithrombin is potentiated by both UFH and LMWHs. FFP will not correct the anticoagulant eect of either UFH or LMWH and may, in fact, worsen the problem.

MANAGEMENT OF POSTOPERATIVE BLEEDING

1205

FFP may be used to correct aPTT prolongations caused by direct thrombin inhibitors such as hirulog and lepirudin, with one important caveat. Circulating, unbound inhibitor binds the prothrombin in the FFP. For this reason, the eective dose is diminished. The amount of FFP needed to normalize aPTT will be greater than might otherwise be estimated to correct a simple factor deciency, for example [1]. Chronic illness, malignancy, malnutrition, or treatment with broad-spectrum antibiotics may result in the rapid onset of vitamin K deciency through reduction in the vitamin Kdependent coagulation factors VII, IX, X, and II. Vitamin K deciency prolongs both the PT and the aPTT. This problem is prevented by the routine administration of vitamin K weekly or biweekly to vulnerable patients. Warfarin sodium has an eect that parallels to vitamin K deciency. Prolongation of PT is used to monitor adequacy of treatment. An INR of 1.5 or less generally allows satisfactory hemostasis during and after surgery, except for intracranial procedures and other critical situations in which any increased risk is problematic. PT prolongations are corrected by the oral administration of vitamin K in doses of 1 to 5 mg. Sometimes multiple doses must be given over time [14] because of warfarins long half-life (about 40 hours), during which the reversal wears o. Complicating factors in warfarin use include malabsorption, hepatic disease, and the simultaneous administration of other anticoagulants. Principles in the management of the patient who has increased INR are as follows [1]: 1. For prolonged INR less than 5.0 without bleeding, hold the next dose 2. For INR between 5.0 and 9.0 without signicant bleeding either a. Withhold next 1 or 2 doses if clinically reasonable or b. Withhold next dose and administer vitamin K, 1.0 to 2 mg orally 3. For INR between 5.0 and 9.0 when rapid normalization is required, administer vitamin K, 2.0 to 4.0 mg, for reduction of INR within 24 hours 4. For INR above 9.0 a. Without signicant bleeding, administer vitamin K, 3.0 to 5.0 mg b. With serious bleeding, administer vitamin K, 10 mg intravenously, and FFP; monitor INR at 12-hour intervals because additional vitamin K supplementation may be required c. With life-threatening hemorrhage, consider prothrombin complexes in addition to measures in (b) 5. For INR above 20.0 a. Whether or not serious bleeding is present, administer vitamin K, 10 mg intravenously, and FFP to reduce INR to therapeutic levels; monitor INR at 12-hour intervals because additional vitamin K supplementation may be required b. With life-threatening hemorrhage, consider administration of prothrombin complexes in addition to measures in (a)

1206

DAGI

Thrombolytic agents. Acute vascular thrombosis in the perioperative setting may be treated with thrombolytic agents such as streptokinase or urokinase that act as plasminogen activators, or with tissue plasminogen activators such as Alteplase, recombinant and Reteplase. Free plasmin degrades both pathologic and physiologic brin clots of recent origin, as well as lysing brinogen and factors V and VIII. These drugs generate large amounts of plasmin [21]. Plasmin has other eects on hemostasis as well, the sum total of which is described broadly as chaotic [14]. The eects of these thrombolytic agents are reversed with anti-brinolytics such as e-aminocaproic acid and tranexamic acid, or by aprotinin, a protease inhibitor. FFP increases levels of factor V and of a2-antiplasmin, a physiologic plasmin inhibitor. Cryoprecipitate increases brinogen and factor VIII levels and is indicated for use in the presence of active bleeding, reduction of thrombin-clottable brinogen below 70 mg/dL, of factor V below 35%, and of factor VIII below 30%. Conversely, assessments of these factors help sort out the diagnosis.

Diuse intravascular coagulation DIC is a complex condition with both thrombotic and hemorrhagic features. It may occur whenever the coagulation system is activated and the level of thrombin increases. Thrombin promotes coagulation, also activates anticoagulant pathways and modulates brinolysis. The severity of the coagulopathy depends on brinogen levels and reects the balance between brinogen production and physiological antithrombotic mechanisms. The imbalance between thrombosis and brinolysis creates a complex hemostatic disorder with both thrombotic and thrombolytic features. The thrombolytic features are responsible for bleeding diatheses. In mild forms of DIC, microthrombi are formed in the vascular system and are cleared. The condition may remain subclinical. In moderate forms, microthrombi occlude the circulation and may be associated with acute respiratory distress syndrome or hepatorenal failure. The severe form more familiar to surgeons occurs when massive brinolysis intervenes to clear the thrombi. Bleeding follows at sites of endothelial damage when the clots are lysed. Tissue factor release induces factor VII activation with heightened coagulation. Additional microthrombi form. The cycle continues and eventually depletes or consumes the store of coagulation factors. Hence, DIC is categorized as a consumption coagulopathy [1]. Acute DIC results from an overwhelming and sudden clotting stimulus, such as a massive crush injury, or from a more moderate clotting stimulus accompanied by shock, as in sepsis or abruptio placentae. A slower, chronic, low-level DIC may occur in other circumstances such as fetal death with retained products of conception, cancer, or large aortic aneurysm [1,14,20].

MANAGEMENT OF POSTOPERATIVE BLEEDING

1207

The mechanisms by which septic shock induces DIC are well described in standard textbooks and are not repeated here. DIC may result from acute hepatic insult, such as shock liver, massive hypotension-induced hepatic necrosis, massive hepatic trauma, acute massive hepatitis, or transplantation. It may also result from events in the setting of chronic liver failure ranging from sepsis to peritoneovenous shunt placement. Hepatic disease induces coagulopathy through reduction and abnormality of coagulation factors, prolonged circulating plasmin levels, reduced brinolysis, disordered platelet function, and sometimes through an associated thrombocytopenia. DIC in the obstetric population is associated with abruptio placentae, amniotic uid embolism, septic abortion, and, less commonly, with second-trimester saline-induced abortions. The low-grade DIC associated with retained products of conception may not manifest until 4 to 5 weeks after death of the fetus. Malignancies, particularly slower-growing adenocarcinomas, are associated with a low-grade and chronic DIC. Chemotherapy, radiotherapy, and other modalities that induce tumor necrosis may accelerate the process. This problem may become more common as tumor angiogenesis inhibitors come to market. A coagulopathy with mixed thrombotic/hemorrhagic features may result in bleeding from acute promyelocytic leukemia. It is accompanied by hypobrinogenemia, increased brin degradation, and thrombocytopenia, possibly ascribable to platelet consumption. Blood products to correct the coagulopathy should be given before surgical intervention. In trauma, DIC is common, particularly in the settings of head injury, multiple fractures with fat embolism, extensive soft tissue damage, and burns over large areas. Potential comorbidities that should be anticipated include shock, hypothermia, acidosis, sepsis, adult respiratory distress syndrome, adrenal failure, massive occult hemorrhage, and sepsis. DIC has been reported after volume expansion with hydroxyethyl starch (hetastarch). The likelihood of bleeding correlates with age, duration of treatment, and dose (O5 mL/kg). The likelihood increases in the presence of renal failure and other underlying hemostatic defects [14,22,23]. DIC is associated with the administration of specic blood products under particular circumstances. These include the rapid transfusion of prothrombin complex concentrates to correct coagulopathy in patients who have liver disease, after warfarin administration, or after the administration of factor VII to correct deciencies. The problem has been avoided by the addition of small concentrations of heparin to the mixture and slower administration. DIC has also been associated with recombinant factor VII administration. A doseresponse eect has been reported [14]. Finally, the chronic low-grade DIC associated with vascular malformations and slowly expanding aortic aneurysms can transform into an acute DIC.

1208

DAGI

Principles of management in acute diuse intravascular coagulation The following principles guide the management of acute DIC [14]: 1. Diagnosis and treatment of the underlying precipitating factors respon sible for the clotting stimulus (may require debridement and amputation of devitalized tissue, drainage of abscesses, or other, similar measures) 2. Cryoprecipitate infusion when brinogen levels are below 70 mg/dL and factor VIII-C is below 40% 3. FFP plasma when factor V levels are below 40% 4. Platelet transfusion to raise platelet count above 50,000 or 100,000, depending on the underlying clinical situation and the type of surgery performed (eg, above 100,000 after intracranial surgery) 5. Low doses of heparin if thrombin activation persists 6. Continuing to work toward normalization of the INR, replacement of blood volume, and restoration of hematocrit In chronic DIC associated with malignancy, patients generally respond to low levels of heparin while undergoing chemotherapy. In patients who have liver disease, FFP provides replacement of coagulation factors. If volume becomes a problem, diuresis or even plasmapharesis may be indicated. In the face of demonstrated excessive brinolysis without DIC, brinolysis inhibitors are recommended. In patients undergoing repair of abdominal aneurysm, factor replacement usually suces, although other stratagems are used as well [14]. After trauma, the emphasis is on volume replacement, treatment of sepsis, and blood product replacement as called for by assessment of clinical status and circulating coagulation factors.

Renal dysfunction and uremia Renal failure induces platelet dysfunction. Some antibiotics and anti-inammatory agents exacerbate this situation. The simplest diagnostic clinical assay is the bleeding time, which should be reserved, however, for patients who have adequate platelet counts (O60,000). For prolonged bleeding times in the presence of abnormal bleeding, the treatment options include [14] 1. Desmopressin acetate at 0.3 mg/kg given intravenously over 20 to 30 minutes (second or third doses may be given, but tachyphylaxis is to be expected if the interval between doses is less than 3 to 5 days) 2. Transfusion of fresh platelets after dialysis 3. Conjugated estrogens at 0.6 mg/kg intravenously each day for 5 days (eect lasts about 2 weeks) 4. Correction of anemia (hematocrit O 26%) with erythropoietin The use of cryoprecipitate is controversial [14].

MANAGEMENT OF POSTOPERATIVE BLEEDING

1209

Uterine bleeding Excessive uterine bleeding has been reported after minor surgical interventions, such as the placement of an intrauterine device or cone biopsy of the cervix, and after other uterine surgery in which no reason for hemorrhage could be found. Uterine bleeding of this nature has been ascribed to localized hyperbrinolysis and to high uterine plasminogen activator activity. It has been linked statistically to the secretory phase of the menstrual cycle and to menstrual bleeding at the time of surgery. The bleeding responds readily to antibrinolytic agents including intravenous or oral e-aminocaproic acid [14]. Postprostatectomy hemorrhage Urinary urokinase that seeps into the surgical bed after prostatectomy may lead to increased local plasmin production and an anticoagulant eect attributable to local hyperbrinolysis. The mechanism is similar to that ascribed to unanticipated uterine bleeding. Antibrinolytic agents successfully reduce bleeding in this setting as well. Bleeding responds to the administration of intravenous or oral e-aminocaproic acid [14]. Posttransfusion coagulopathy Coagulopathy may follow after the rapid replacement of two thirds of the bodys blood volume. This level of transfusion is generally required only after massive trauma, although it may be called for in the course of vascular and oncologic surgery. The roots of this problem lie in the changes that occur when whole blood is stored. After several days, the concentration of two thermolabile coagulation factors, V and VIII-C, is reduced by 15% to 50%. After 2 or 3 days, platelet function is lost. Once transfused, platelets that do not function well are rapidly removed from circulation. As a result, bleeding time may be prolonged irrespective of the platelet count. A consumptive coagulopathy ensues frequently resulting in thrombocytopenia. The scenario often includes massive tissue injury, hypotension, acidosis, hypoxia, some level of hepatorenal dysfunction, and hypothermia, all of which contribute to hemostatic defects and may result in DIC. McKenna [14], however, asserts that overtly decompensated DIC is uncommon in this setting [1]. Whole blood has the advantage of availability and volume. For the reasons outlined previously, however, transfusion with components of whole blood is theoretically preferable to whole blood transfusion. One protocol recommends transfusing one unit of FFP for every three to four units of

1210

DAGI

RBC when patients are expected to lose more than one unit of blood volume rapidly [14]. Platelet concentrates are used to correct bleeding time and treat thrombocytopenia. The details of and controversies surrounding resuscitation theory are beyond the scope of this article, but the principle of choosing blood components over whole blood when possible and of anticipating a multifactorial coagulopathy in the setting of massive transfusion is broadly accepted [1]. Treatment should involve replacement of diluted factors by FFP, cryoprecipitate, calcium, and platelets. Replacement is continued until coagulation parameters normalize. Bleeding from non-technical causes should then cease. Posttransfusion purpura syndrome Posttransfusion purpura syndrome refers to the sudden onset of severe thrombocytopenia 7 to 10 days after transfusion of RBC, FFP, or platelets. The thrombocytopenia is accompanied by sudden and dramatic hemorrhage. Mortality from intracranial hemorrhage is high. The dierential diagnosis includes sepsis and drug-induced thrombocytopenia. The population at greatest risk seems to be multiparous women. Sensitization is generally attributed to pregnancy or, less commonly, to previous transfusions of blood products. The syndrome is a form of transfusion reaction. It occurs when the human platelet antigen (HPA-1a) antigen in blood products elicits an antiHPA-1a antibody in the 2.5% of whites and 0.5% of African Americans who lack the antigen. The precise mechanisms that lead to platelet destruction are not understood. The management of posttransfusion purpura syndrome involves urgent platelet transfusion after the documentation of posttransfusion purpura syndrome antibodies in a serum sample. This treatment may seem counterintuitive, but the intention is to deplete the store of available anti-HPA1a platelet antibodies through the destruction of newly supplied HPA1a-positive platelets while simultaneously depending on the non-antigenic platelets and on platelet fragments to assist in hemostasis and control the bleeding. Plasmapharesis scrubs the antibodies in 80% of the cases, and intravenous IgG blocks further antigenantibody reactions. Steroids may be useful in nonspecic indications such as vascular support and treatment of overall stress [14]. Hemophilia Hemophilia rarely enters the dierential diagnosis in the absence of a personal or family history of bleeding. There are three common forms. Type A involves a deciency of factor VIII, type B involves a deciency of factor IX, and type C involves a deciency of factor XI. The diagnosis is made through specic factor analysis. Treatment involves replacement of the factor or factors that are decient. After extensive

MANAGEMENT OF POSTOPERATIVE BLEEDING

1211

transfusions, hemophiliac patients may develop antibodies to blood products or to bioengineered factors of recombinant origin, such as recombinant activated factor VII. There are approximately 20 approved products on the market providing some combination of factor VIII replacement, factor IX replacement, or von Willebrands factor replacement. Approximately 20% are recombinant, 5% are porcine derived, and the rest are plasma derived. Each of the six products supplying von Willebrands factor also provides factor VIII replacement. Most surgeons will want to seek consultation to oversee the treatment of congenital or familial blood dyscrasias and provide longterm follow-up.

Summary The risks and complications associated with postoperative bleeding can be substantially reduced through measures such as adequate preoperative assessment; the identication and correction of deciencies in circulating clotting factors, in platelet number and function, in hematocrit and blood volume, and body temperature; and the treatment of infection. It is important to optimize the condition of the patient before surgery. A riskbenet calculus is always at play. If surgery can be delayed until all pertinent problems are identied and addressed, so much the better. If not, the patients risk may be reduced by choosing a less invasive or less radical procedure, by appropriate preoperative consultations, by staging the operation, by judicious choice of anesthetic technique, by operating on an inpatient rather than an outpatient basis, and by arranging in advance for adequate postoperative monitoring and the appropriate intensity of care. This article has addressed the prevention and management of postoperative hemorrhage. For reasons of focus and space, it has not addressed two important and related subjects: the management of chronic anticoagulation and the problem of prophylaxis for deep vein thrombosis in the perioperative period. Whenever possible, the patient should go to surgery with normal coagulation parameters. The decision to proceed despite abnormal coagulation parameters (within therapeutic range or not) is equivalent to deciding either that the risk of intraoperative and postoperative hemorrhage is of lesser consequence to the patients overall welfare than a delay, or that no marginal benet will accrue from normalizing coagulation. This decision may well be in the patients best interests. The consent obtained for surgery ought to reect in detail the thinking and discussion that surrounded the decision.

Acknowledgments Dr. Rodney Falk provided important insights and perspective.

1212

DAGI

Appendix 1 Prothrombin time and international normalized ratio The traditional method of determining the ecacy of anticoagulation therapy is the prothrombin time (PT). A blood sample is collected in a tube containing citrated sodium. After centrifugation, a specic volume of thromboplastin reagent is added. The interval until a brin clot forms, measured in seconds, is reported as the PT. The thromboplastin reagent can be either an extract of mammalian tissue (lungs heart or brain of animals) rich in tissue factor or a recombinant proportion of human tissue factor in combination with phospholipids. The sensitivity of individual thromboplastin reagents is variable. Higher sensitivity results in prolonged PT. Sensitivity is also aected by shelf life. The international normalized ratio (INR) corrects for potential variability by calibrating results against a standardized measure of the sensitivity of the thromboplastin reagents used. It has become a standard for monitoring oral anticoagulant therapy and serves the same purpose as the PT. The prothrombin time ratio relates the observed PT (in seconds) to each laboratorys calculated mean normal PT (in seconds). Target anticoagulation ranges are commonly expressed in terms of INRs: while a range of 2.0 to 3.0 is recommended for most indications, a range of 2.5 to 3.5 is recommended for patients at higher risk (eg, those with mechanical prosthetic heart valves). The INR alone is insucient to assess coagulopathy. Hirsh and Poller [24] provide a fuller discussion.

References
[1] Owings JT, Gosselin RC. Bleeding and transfusion. In: Wilmore WW, Cheung LY, Harken AH, et al, editors. American College of Surgeons surgery principles and practice. Chicago: American College of Surgeons; 2001. p. 7790. [2] Comprehensive accreditation manual for hospitals: the ocial handbook. Oakbrook Terrace (IL): Joint Commission on the Accreditation of Healthcare Organizations; 1998. [3] Litaker D. Preoperative screening. Med Clin North Am 1999;83:156581. [4] Smetana GW, Macpherson DS. The case against routine preoperative laboratory testing. Med Clin North Am 2003;87:740. [5] Turnbull JM, Buck C. The value of pre-operative screening investigations in otherwise healthy individuals. Arch Intern Med 1987;147:11015. [6] Roizen MF. More preoperative assessment by physicians and less by laboratory tests. N Engl J Med 2000;342:2045. [7] Charpak Y, Blery C, Chastang C, et al. Usefulness of selectively ordered preoperative tests. Med Care 1988;26:95104. [8] Macpherson DS, Snow R, Lofren RP. Preoperative screening; value of previous tests. Ann Intern Med 1990;113:96973. [9] Dzankic S, Pastor D, Gonzales C, et al. The prevalence and predictive value of abnormal preoperative laboratory tests in elderly surgical patients. Anesth Analg 2001;93:3018.

MANAGEMENT OF POSTOPERATIVE BLEEDING

1213

[10] Sanders DP, McKinney FW, Harris WH. Clinical evaluation and cost eectiveness of preoperative laboratory assessment on patients undergoing total hip arthroplasty. Orthopedics 1989;12:144953. [11] Faris PM, Spence RK, Larhol KM, et al. The predictive power of baseline hemoglobin for transfusion risk in surgery patients. Orthopedics 1999;22(Suppl):S13540. [12] Rohrer M, Mechelotti M, Nahrwold D. A prospective evaluation of the ecacy of preoperative coagulation testing. Ann Surg 1988;208:5547. [13] Cobas M. Preoperative assessment of coagulation disorders. Int Anesthesiol Clin 2001;39: 115. [14] McKenna R. Abnormal coagulation in the postoperative period contributing to excessive bleeding. Med Clin North Am 2001;85:1277310. [15] Kereiakis DJ, Berkowitz SD, Linco AM, et al. Clinical correlates and course of thrombocytopenia during percutaneous coronary intervention in the era of abciximab platelet glycoprotein IIb/IIIa blockade. Am Heart J 2000;14:7480. [16] Bennet CL, Connors JM, Carwille JM, et al. Thrombotic thrombocytopenic purpura associated with clopidogrel. N Engl J Med 2000;342:17737. [17] Bellan L. Preoperative testing for cataract surgery. Can J Ophthalmol 1994;29:1114. [18] Simon TL, Alverson DC, AuBuchon J, et al. Practice parameters for the use of red blood cell transfusions: developed by the Red Blood Cell Administration Practice Guideline Development Task Force of the College of American Pathologists. Arch Pathol Lab Med 1998;122: 1308. [19] Despotis GJ, Levine V, Saleem R, et al. Use of point-of-care test in identication of patients who can benet from desmopressin during cardiac surgery: a randomized controlled trial. Lancet 1999;354:10610. [20] Warkentin TE, Levine MN, Hirsch J, et al. Heparin-induced thrombocytopenia in patients treated with low-molecular-weight heparin or unfractionated heparin. N Engl J Med 1995; 332:13305. [21] McKenna R, Walenga JM. Alpha 2-plasma inhibitor (a2-PI) deciency. E Medicine Journal (serial online) 2001. Available from: http://www.emedicine.com/medicine,obstetrics and gynecology, psychiatry and surgery/109.htm. Accessed September 2, 2005. [22] Strauss RG, Stump DC, Henriksen RA. Hydroxyethyl starch accentuates von Willebrands disease. Transfusion 1985;25:2347. [23] Symington BE. Hetastarch and bleeding complications. Ann Intern Med 1986;105:6278. [24] Hirsh J, Poller L. International normalized ratio: a guide to understanding and correcting its problems. Arch Int Med 1994;154:2828.